lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20230420101617.142225-1-krzysztof.kozlowski@linaro.org>
Date:   Thu, 20 Apr 2023 12:16:11 +0200
From:   Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>
To:     Vinod Koul <vkoul@...nel.org>,
        Bard Liao <yung-chuan.liao@...ux.intel.com>,
        Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>,
        Sanyog Kale <sanyog.r.kale@...el.com>,
        Andy Gross <agross@...nel.org>,
        Bjorn Andersson <andersson@...nel.org>,
        Konrad Dybcio <konrad.dybcio@...aro.org>,
        Liam Girdwood <lgirdwood@...il.com>,
        Mark Brown <broonie@...nel.org>,
        Jaroslav Kysela <perex@...ex.cz>,
        Takashi Iwai <tiwai@...e.com>, alsa-devel@...a-project.org,
        linux-kernel@...r.kernel.org, linux-arm-msm@...r.kernel.org
Cc:     Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>,
        Patrick Lai <quic_plai@...cinc.com>
Subject: [PATCH 0/6] ASoC/soundwire: qcom: correctly probe devices after link init

Hi,

Dependencies
============
1. ASoC codec: changes are independent, however they should rather come the same
   cycle as Soundwire changes, to avoid new warning and small delay.

2. Soundwire: changes (context) depend on:
   https://lore.kernel.org/r/20230209131336.18252-3-srinivas.kandagatla@linaro.org
   https://lore.kernel.org/all/20230418095447.577001-1-krzysztof.kozlowski@linaro.org/

Problems solved
===============
Soundwire devices are supposed to be kept in reset state (powered off)
till their probe() or component bind() callbacks.  However if they are
already powered on, then they might enumerate before the master
initializes bus in qcom_swrm_init() leading to occasional errors like:

  qcom-soundwire 6d30000.soundwire-controller: Qualcomm Soundwire controller v2.0.0 Registered
  wcd938x_codec audio-codec: bound sdw:0:0217:010d:00:4 (ops wcd938x_sdw_component_ops)
  wcd938x_codec audio-codec: bound sdw:0:0217:010d:00:3 (ops wcd938x_sdw_component_ops)
  qcom-soundwire 6ad0000.soundwire-controller: swrm_wait_for_wr_fifo_avail err write overflow

The problem primarily lies in Qualcomm Soundwire controller probe() sequence:
1. request_threaded_irq()
2. sdw_bus_master_add() - which will cause probe() and component bind()
   of Soundwire devices, e.g. WCD938x codec drivers.  Device drivers
   might already start accessing their registers.
3. qcom_swrm_init() - which initializes the link/bus and enables
   interrupts.

Any access to device registers at (2) above, will fail because link/bus
is not yet initialized.

Cc: Patrick Lai <quic_plai@...cinc.com>

Best regards,
Krzysztof

Dmitry Torokhov (1):
  ASoC: wcd938x: switch to using gpiod API

Krzysztof Kozlowski (5):
  ASoC: codecs: wcd938x: Keep device in reset till bind
  ASoC: codecs: wcd938x: Check for enumeration before using TX device
  soundwire: qcom: drop unused struct qcom_swrm_ctrl members
  soudnwire: master: protect concurrecnt check for bus->md
  soundwire: qcom: do not probe devices before bus/link init

 drivers/soundwire/master.c |  7 ++-
 drivers/soundwire/qcom.c   | 92 +++++++++++++++++++++++++++++---------
 sound/soc/codecs/wcd938x.c | 44 +++++++++++-------
 3 files changed, 107 insertions(+), 36 deletions(-)

-- 
2.34.1

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ